Coded Computation Against Distributed Straggling Channel Decoders in the Cloud for Gaussian Uplink Channels

TL;DR

This paper investigates a cloud-based decoding scheme for Gaussian uplink channels in CRAN, using linear coding to mitigate straggler effects, and derives bounds on error rate and latency.

Contribution

It introduces a linear coding approach at the cloud to counteract straggling decoders and provides analytical bounds on FER versus latency for Gaussian channels.

Findings

Maximum user rate for reliable decoding is derived.

Two upper bounds on frame error rate as a function of latency are developed.

The approach effectively mitigates straggler impact in cloud decoding.

Abstract

The uplink of a Cloud Radio Access Network (CRAN) architecture is studied, where decoding at the cloud takes place at distributed decoding processors. To mitigate the impact of straggling decoders in the cloud, the cloud re-encodes the received frames via a linear code before distributing them to the decoding processors. Focusing on Gaussian channels, and assuming the use of lattice codes at the users, in this paper the maximum user rate is derived such that all the servers can reliably recover the linear combinations of the messages corresponding to the employed linear code at the cloud. Furthermore, two analytical upper bounds on the frame error rate (FER) as a function of the decoding latency are developed, in order to quantify the performance of the cloud's linear code in terms of the tradeoff between FER and decoding latency at the cloud.